Item14420: filter out non-breaking zero-width space as well as the br…

…eaking version. TinyMCE 4 has started using it, and still doesn't filter it from output. Also rename a couple of functions in Node to reflect modern HTML usage.
foswiki · Aug 26, 2017 · cda2014 · cda2014
1 parent 2d1a826
commit cda2014
Show file tree

Hide file tree

Showing 2 changed files with 45 additions and 39 deletions.
diff --git a/WysiwygPlugin/lib/Foswiki/Plugins/WysiwygPlugin/HTML2TML.pm b/WysiwygPlugin/lib/Foswiki/Plugins/WysiwygPlugin/HTML2TML.pm
@@ -196,13 +196,18 @@ DEFAULT
     $text =~ s/\&\#x22;/\&quot;/goi;
     $text =~ s/\&\#160;/\&nbsp;/goi;
 
-    # SMELL: Item11912 These are left behind by TMCE as zero width characters
-    # surrounding italics and bold inserted by Ctrl-i and Ctrl-b
-    # We really ought to have a better solution.  TMCE is supposed
-    # to handle this it the cleanup routine, but it doesn't happen,
-    # and cleanup routine has been deprecated.
-    $text =~ s/&#xFEFF;//g;    # TMCE 3.5.x
-    $text =~ s/&#x200B;//g;    # TMCE pre 3.5
+    # Item11912 Item14420 Zero-width space (x200B) and Non-breaking
+    # zero-width space (xFEFF) are left behind by TinyMCE with
+    # italics and bold inserted by Ctrl-i and Ctrl-b. This is probably
+    # a TinyMCE bug, but in general these characters are useless in TML
+    # so we strip them in all their forms anyway.
+    foreach my $d ( 0xFEFF, 0x200B ) {
+        $text =~ s/&#$d;//g;    # decimal entity
+        my $c = chr($d);
+        $text =~ s/$c//g;
+        my $h = sprintf( "%04x", $d );
+        $text =~ s/&#x$h;//g;    # hex entity
+    }
 
     HTML::Entities::_decode_entities( $text, safeEntities() );
 

diff --git a/WysiwygPlugin/lib/Foswiki/Plugins/WysiwygPlugin/HTML2TML/Node.pm b/WysiwygPlugin/lib/Foswiki/Plugins/WysiwygPlugin/HTML2TML/Node.pm
@@ -1535,36 +1535,7 @@ sub _handleABBR    { return _flatten(@_); }
 sub _handleACRONYM { return _flatten(@_); }
 sub _handleADDRESS { return _flatten(@_); }
 
-sub _handleB {
-    my ( $this, $options ) = @_;
-    if ( $options & IN_TABLE ) {
-        if (
-            $this->{parent}
-            && (   $this->{parent}->{tag} eq 'td'
-                || $this->{parent}->{tag} eq 'th' )
-          )
-        {
-            # Item9651: Don't convert bold in a table cell into stars
-            # if the TML would be interpreted as a heading
-            my $left  = $this->{prev} ? $this->{prev}->stringify() : '';
-            my $right = $this->{next} ? $this->{next}->stringify() : '';
-            if ( "$left$right" =~ /^\s*$/ ) {
-
-                # Have to exclude ==this case== and __that case__ as
-                # they look awfully similar but don't generate table
-                # headings
-                my ( $foo, $t ) = _emphasis( @_, '*' );
-                if ( $t !~ /.*?$CHECK1(==|__).*\1$CHECK2/ ) {
-                    return ( 0, undef );
-                }
-                else {
-                    return ( $foo, $t );
-                }
-            }
-        }
-    }
-    return _emphasis( @_, '*' );
-}
+sub _handleB        { return _handleSTRONG(@_); }
 sub _handleBASE     { return ( 0, '' ); }
 sub _handleBASEFONT { return ( 0, '' ); }
 
@@ -1723,7 +1694,7 @@ sub _handleH3   { return _H( @_, 3 ); }
 sub _handleH4   { return _H( @_, 4 ); }
 sub _handleH5   { return _H( @_, 5 ); }
 sub _handleH6   { return _H( @_, 6 ); }
-sub _handleI    { return _emphasis( @_, '_' ); }
+sub _handleI    { return _handleEM(@_); }
 
 sub _handleIMG {
     my ( $this, $options ) = @_;
@@ -1964,7 +1935,37 @@ sub _handleSPAN {
 
 # STRIKE
 
-sub _handleSTRONG { return _emphasis( @_, '*' ); }
+sub _handleSTRONG {
+    my ( $this, $options ) = @_;
+
+    if ( $options & IN_TABLE ) {
+        if (
+            $this->{parent}
+            && (   $this->{parent}->{tag} eq 'td'
+                || $this->{parent}->{tag} eq 'th' )
+          )
+        {
+            # Item9651: Don't convert bold/strong in a table cell into stars
+            # if the TML would be interpreted as a heading
+            my $left  = $this->{prev} ? $this->{prev}->stringify() : '';
+            my $right = $this->{next} ? $this->{next}->stringify() : '';
+            if ( "$left$right" =~ /^\s*$/ ) {
+
+                # Have to exclude ==this case== and __that case__ as
+                # they look awfully similar but don't generate table
+                # headings
+                my ( $foo, $t ) = _emphasis( @_, '*' );
+                if ( $t && $t !~ /.*?$CHECK1(==|__).*\1$CHECK2/ ) {
+                    return ( 0, undef );
+                }
+                else {
+                    return ( $foo, $t );
+                }
+            }
+        }
+    }
+    return _emphasis( @_, '*' );
+}
 
 sub _handleSTYLE { return ( 0, '' ); }